Skip to main content

A plugin to leverage ML segmentation in napari

Project description

Allencell-segmenter-ml

Test and lint

What is Allen Cell Segmenter ML

A deep learning-based segmentation Napari plugin to curate datasets, train your own model (UNET), and run inference on 2D and 3D cell data.

📰 News

  • [2024.09.24] :tada: Initial release of the plugin and Megaseg models!

:hammer_and_wrench: Installation

System Requirements

We currently support Windows, MacOS, and Linux operating systems. The minimum system requirements are:

  • 8GB of RAM
  • 8 CPU Cores
  • 1 NVIDIA GPU with 8GB of VRAM (optional)

NOTE: If you plan to use the plugin without a GPU, training will default to using your CPU and will be significantly slower. A GPU is highly recommended for training models. Depending on how large your images are---2D vs 3D, resolution, model size---running inference may also be slow without a GPU.

Pre-Installation

STEP 1. Install Python

Before installing the plugin, please make sure you have the following installed:

  • Python 3.10 or later

New to Python? We recommend installing Python 3.10 through the official Python website. This will include the pip package manager, which is required to install the plugin.

If you are unsure if you have Python installed or which version you may have, you can check by running the following command in your terminal or powershell:

# Check version of python
python --version

# If the above does not work, try this one
python3 --version

# Specifically check for Python 3.10
python3.10 --version
STEP 2. Create a Virtual Environment

Next we will create a new Python environment to install the plugin. This will help avoid conflicts with other packages you may have installed by creating an isolated environment for the plugin to live in. In general, it is good practice to choose a name for your environment that is related to either the project you are working on or the software you are installing. In this case, we use venv-allen-segmenter-ml where venv stands for virtual environment.

Navigate to where you want to create a new environment (Example. Documents), run the following command in your terminal or powershell:

# Create a new environment
python3.10 -m venv venv-allen-segmenter-ml

# Activate the environment
source venv-allen-segmenter-ml/bin/activate

Confirm Virtual Environment is Activated

To confirm that the virtual environment has been successfully activated, you can follow these steps:

  1. Check that the prompt includes the name of your virtual environment, venv-allen-segmenter-ml. It should look something like this:

    (venv-allen-segmenter-ml) $
    
    # Example on a Windows machine
    (venv-allen-segmenter-ml) PS C:\Users\Administrator\Documents> 
    
  2. Run the following command to verify Python 3.10 is being used within the virtual environment:

    python --version
    
    # Python 3.10.11   <-- Example output
    

Install the Plugin

To install the latest version of the plugin:

pip install allencell-segmenter-ml

:rotating_light: Post-Installation :rotating_light:

:memo: NOTE: This section is specifically for users with at least one NVIDIA GPU installed on their machine. Not sure if you have an NVIDIA GPU? You can check by running nvidia-smi as shown below. If you do not have an NVIDIA GPU system, you can skip this section.

Required Package

After installing the plugin, you need to install a PyTorch version that is compatible with your system. PyTorch is a deep learning library that is used to train and run the models in the plugin. We understand that everyone manages CUDA drivers and PyTorch versions differently depending on their system and use cases, and we want to respect those decisions because CUDA drivers can be a pain.

STEP 1. Checking CUDA Version

To check your CUDA version, you can run the following command in your terminal or powershell:

nvidia-smi

As an example, the output will look similar to this. My CUDA Version is 11.8:

PS C:\Users\Administrator> nvidia-smi
Fri Sep 13 03:22:15 2024
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 522.06       Driver Version: 522.06       CUDA Version: 11.8     |
|-------------------------------+----------------------+----------------------+
| GPU  Name            TCC/WDDM | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  Tesla T4           TCC   | 00000000:00:1E.0 Off |                    0 |
| N/A   27C    P8     9W /  70W |      0MiB / 15360MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

STEP 2. PyTorch Installation

To Install PyTorch, please visit the PyTorch website and select the appropriate installation options for your system. An example is provided below.

torch-install

PyTorch Installation for Windows, MacOS, and Linux

Example

For instance, if I am using

  • Windows workstation
  • pip package manager
  • Python (3.10)
  • CUDA 11.8

Then the command for me would be:

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

If the installation is successful, let's test just to be sure that your GPU is detected by PyTorch. Run the following command in your terminal or powershell:

python -c "import torch; print(torch.cuda.is_available())"

You should see True if your GPU is detected (see below). If you see False, then PyTorch is not detecting your GPU. You may need to reinstall PyTorch or check your CUDA drivers. Double check that your virtual environement is activated (venv-allen-segmenter-ml).

(venv-allen-segmenter-ml) PS C:\Users\Administrator\Documents> python -c "import torch; print(torch.cuda.is_available())"
True

:tada: You have successfully installed the plugin and PyTorch. You are now ready to use the plugin!


Running the Plugin

To run the plugin (and verify the installation), you can use the following command in your terminal or powershell:

napari

You should see the below window pop up. To start using the plugin, click on the Plugins tab and select Allen Cell Segmenter ML:

plugin

Allen Cell Segmenter ML Launching the Plugin.

Models

Model Model Name Available in Plugin Model Size (MB) Supported Magnifications
MegaSeg-S megaseg_light 4.8MB 100X
MegaSeg-M megaseg_medium Coming soon! TBD TBD
MegaSeg-L megaseg_large 191MB 20X, 40X, 67X, 100X

Download and using the Megaseg Models

To use the MegaSeg models in the plugin, you can download them from the dropdown menu shown below:

download-model

Download the MegaSeg Model for use in the Plugin

A popup window will appear and you can select which model you would like to download. Once the download is complete, another popup will let you know the download was successful and where the model was downloaded.

select-megaseg

Select the MegaSeg model to Run Inference

To use the model for inference on your images, choose Select an existing model, select the megaseg model you downloaded, and click Apply. You can now use the model to segment your images!

License

Distributed under the terms of the Allen Institute Software License license.

Issues

If you encounter any problems, please file an issue along with a detailed description.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

allencell_segmenter_ml-0.1.16rc0.tar.gz (4.2 MB view details)

Uploaded Source

Built Distribution

allencell_segmenter_ml-0.1.16rc0-py3-none-any.whl (243.6 kB view details)

Uploaded Python 3

File details

Details for the file allencell_segmenter_ml-0.1.16rc0.tar.gz.

File metadata

File hashes

Hashes for allencell_segmenter_ml-0.1.16rc0.tar.gz
Algorithm Hash digest
SHA256 aacbd8690dd42a9670bbd1662528ac482d5e1b787f3fb282cffbadf9a1f30f27
MD5 2f650347c6f6c1f60fa1b48be41e0f1d
BLAKE2b-256 2617384719458188ec3b8f7acd8fc31cf196c2c85717d6a7e5bb5f3e4f1b7429

See more details on using hashes here.

File details

Details for the file allencell_segmenter_ml-0.1.16rc0-py3-none-any.whl.

File metadata

File hashes

Hashes for allencell_segmenter_ml-0.1.16rc0-py3-none-any.whl
Algorithm Hash digest
SHA256 15e1402300c6fbf47736d2728da7c03f48099cef317ee6b8780a8eadcb5693b6
MD5 ffd9ad9e313cb4276197e424924726f6
BLAKE2b-256 a753fc4b0def059e0c7db684d0e93495a91ebb7b93c6fc8b64c8f8321dd4bfba

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page